Background technology
In mr imaging technique, the speed of imaging weighs a major criterion of formation method.The key factor of restriction image taking speed comprises the speed of data acquisition and k-space filling.General data acquisition modes will adopt full k-space data, then just can carry out reconstruction and obtain image.Magnetic resonance parallel gathers reconstruction technique, and be the mode utilizing coil restructuring to merge, fill up the data of lack sampling, utilization is filled up complete k-space data and rebuild.Profit in such a way, according to demand, can only gather a part of k-space data, need not adopt completely whole k-space.Utilize such method greatly can accelerate the speed of imaging.
One of relatively more conventional method for parallel reconstruction is broad sense self calibration parallel acquisition (GeneralizedAutocalibrating Partially parallel acquisitions, GRAPPA).Fig. 1 is the sample graph of the GRAPPA method of 4 passages (coil 1,2,3,4) matching.With reference to Fig. 1, the method of traditional GRAPPA is: the point of white represents the lack sampling data be not filled, the point representative of grey is just by data that the method for matching is filled, the point of black represents the data of actual samples, in Fig. 1, the point of any one white can be expressed as the linear superposition of the point of surrounding black, be equivalent to merge the data of multiple coil, and merge coefficient n
ijrepresent i-th coil, a jth position, can be determined by the point of the some matching grey of black, the line between the point radiateing other each black in Fig. 1 from the point of certain grey represents the relation of the point of matching grey.Merge coefficient determines that the data of coil can merge according to the merge coefficient of trying to achieve by the point of rear white, and data of plugging a gap.Fig. 2 is the sampling schematic diagram utilizing traditional coil merging method to calculate merge coefficient.With reference to Fig. 2, have 4 passages, i.e. coil 1 ', 2 ', 3 ', 4 ', merge coefficient n
ijthe form being write as vector has following relation:
Wherein,
represent the vector representation of Grey Point,
represent the vector representation of black color dots,
for the vector representation of merge coefficient, arrow 22 represents the data relationship utilized, and the suspension points in figure represents remaining data point, in this coil merging method, is superposed obtain the point of grey by the point Linear of black, utilizes formula (1) can be in the hope of
then slip square frame 21, square frame 21 represents that matching coil merge coefficient needs the data area of reference, and all data in square frame 21 are a data reconstructed block, utilizes the point of black and vector
product can the data point represented by point of white in calculating chart 2, namely can know the data point that k-space lacks.Fig. 3 a fills up without the method for traditional GRAPPA, the image after directly utilizing the k-space of missing data to rebuild; Fig. 3 b is the image rebuild after disappearance k-space data filled up by the method merging coil of traditional GRAPPA.The speedup factor of Fig. 3 a and Fig. 3 b is 2, artifact with reference to Fig. 3 a image is serious, and the method through traditional GRAPPA in comparison diagram 3a, Fig. 3 b eliminates a lot of artifact, but part as shown by arrows, still some convolution artifact (artifact).Therefore, although utilize said method can reduce the k-space data of disappearance to a certain extent, when the signal to noise ratio (S/N ratio) (signal-to-noise ratio, SNR) of data is lower time, formula (1) is utilized to try to achieve
have certain difference with optimum value, if only consider some data reconstruction blocks, this fitting result can only to represent in this square frame best merge coefficient, and differs in other square frame and be decided to be the best.
Summary of the invention
Technical matters to be solved by this invention there is provided a kind of magnetic resonance parallel image acquisition and image rebuilding method, to solve because merge coefficient out of true causes the problem of artifact.
In order to solve the problems of the technologies described above, technical scheme of the present invention is: provide a kind of magnetic resonance parallel image acquisition method, multichannel lack sampling k-space data is used to carry out the full sampled k-space data of matching, described parallel acquisition method comprises: in described lack sampling k-space, set at least two fitting module, the structure of the k-space matrix comprised in each described fitting module is identical, and includes the data of the k-space data of actual acquisition and the k-space data institute matching by described actual acquisition in each described fitting module; Described two or more fitting module is utilized to obtain merge coefficient; Utilize described merge coefficient to calculate the data of lack sampling, fill up described lack sampling k-space, form full sampled k-space.
Further, in described each fitting module, matrix unit arrangement mode is identical.
Further, described two or more fitting module fitting module of comprising the first fitting module and being formed after moving the first fitting module along frequency coding direction.
Further, described merge coefficient, by two or more fitting module described being merged the total fitting module of generation one, calculates to obtain described merge coefficient by described total fitting module.
Further, described merge coefficient calculates gained by carrying out least square method to total fitting module.
Further, the k-space data of each passage is comprised in described fitting module.
The present invention also provides a kind of method for reconstructing of magnetic resonance image (MRI), utilizes above-mentioned magnetic resonance parallel image acquisition method to obtain full sampled k-space data, is converted into the view data of image area, merges the view data of described each passage thus realize image reconstruction.
Magnetic resonance parallel image acquisition method provided by the invention, merge coefficient is obtained by least two fitting module, the data volume of matching is increased, compensate for the impact that signal to noise ratio (S/N ratio) is poor, and fully taken into account each available fitting module, make fitting coefficient reach best whole entirely adopting in space, thus merge coefficient can be obtained more exactly, remove the artifact brought because of merge coefficient out of true efficiently.
The method for reconstructing of magnetic resonance image (MRI) provided by the invention, the full sampled k-space data utilizing magnetic resonance parallel image acquisition method to obtain realize image reconstruction, effectively can remove the artifact of rebuilding in image.
Accompanying drawing explanation
Fig. 1 is the sampling schematic diagram of the GRAPPA method of 4 passages (coil 1,2,3,4) matching;
Fig. 2 is the sampling schematic diagram utilizing traditional coil merging method to calculate merge coefficient;
Fig. 3 a fills up without the method for traditional GRAPPA, the image after directly utilizing the k-space of missing data to rebuild;
Fig. 3 b is the image rebuild after disappearance k-space data filled up by the method merging coil of traditional GRAPPA;
Fig. 3 c utilizes the method for the embodiment of the present invention when speedup factor is 2, the image rebuild after filling up disappearance k-space data;
Fig. 4 is the steps flow chart schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides;
Fig. 5 is the sampling schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides;
Fig. 6 is the easy sampling schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides;
Fig. 7 a to be speedup factor be 4 the method without traditional GRAPPA fill up, the image after directly utilizing the k-space of missing data to rebuild;
Fig. 7 b to be speedup factor be 4 the method through traditional GRAPPA merge the image rebuild after disappearance k-space data filled up by coil;
Fig. 7 c utilizes the method for the embodiment of the present invention when speedup factor is 4, the image rebuild after filling up disappearance k-space data.
Embodiment
A kind of magnetic resonance parallel image acquisition proposed the present invention below in conjunction with the drawings and specific embodiments and image rebuilding method are described in further detail.According to the following describes and claims, advantages and features of the invention will be clearer.It should be noted that, accompanying drawing all adopts the form that simplifies very much and all uses non-ratio accurately, only for object that is convenient, the aid illustration embodiment of the present invention lucidly.
Core concept of the present invention is, magnetic resonance parallel image acquisition method provided by the invention, merge coefficient is obtained by least two fitting module, the data volume of matching is increased, compensate for the impact that signal to noise ratio (S/N ratio) is poor, and fully taken into account each available fitting module, make fitting coefficient reach best whole entirely adopting in space, thus merge coefficient can be obtained more exactly, remove the artifact brought because of merge coefficient out of true efficiently.The method for reconstructing of magnetic resonance image (MRI) provided by the invention, the full sampled k-space data utilizing magnetic resonance parallel image acquisition method to obtain realize image reconstruction, effectively can remove the artifact of rebuilding in image.
Fig. 4 is the steps flow chart schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides, and the invention provides a kind of magnetic resonance parallel image acquisition method, comprising:
S41, in described lack sampling k-space, set at least two fitting module, the structure of the k-space matrix comprised in each described fitting module is identical, and includes the data of the k-space data of actual acquisition and the k-space data institute matching by described actual acquisition in each described fitting module;
S42, described two or more fitting module is utilized to obtain merge coefficient;
S43, utilize described merge coefficient to calculate the data of lack sampling, fill up described lack sampling k-space, form full sampled k-space.
Preferably, described merge coefficient, by two or more fitting module described being merged the total fitting module of generation one, calculates to obtain described merge coefficient by described total fitting module.In the present embodiment, above-mentioned merge coefficient calculates gained by carrying out least square method to total fitting module.
Preferably, in described each fitting module, matrix unit arrangement mode is identical.Matrix unit arrangement mode is identical specifically refers to that the line number of matrix unit in each fitting module is identical with columns.The fitting module that described two or more fitting module comprises the first fitting module and formed after moving the first fitting module along frequency coding direction.
Below in conjunction with sampling schematic diagram, magnetic resonance parallel image acquisition method of the present invention is described in more detail, which show the preferred embodiments of the present invention, should be appreciated that those skilled in the art can revise the present invention described here, and still realize advantageous effects of the present invention.
Fig. 5 is the sampling schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides, 4 passages are had in the sampling schematic diagram of the embodiment of the present invention, represent coil 1 ' respectively, 2 ', 3 ', 4 ', in figure, white point represents lack sampling data to be filled, Grey Point representative is just by data that the method for matching is filled, black color dots represents the data of actual samples, in the data of entirely adopting, set some fitting module, the structure of the k-space matrix comprised in each described fitting module is identical, the structure of k-space matrix is identical refers to that the character of same column matrix unit is identical, namely same column matrix unit is all black color dots or is all Grey Point or is all white point.Black color dots and Grey Point is comprised in each described fitting module, the k-space data of each passage is comprised in fitting module, in embodiments of the present invention, the fitting module of all settings is all included in calculating, but in order to set forth conveniently, in Figure 5 whole fitting module is not showed, represent all the other fitting module do not identified with suspension points.In the present embodiment, only with fitting module 51,52,53 is representative elaboration scheme, fitting module 51 with to comprise matrix unit arrangement mode in fitting module 53 identical, the position of fitting module 53 be the position of fitting module 51 to right translation two column matrix unit, the fit approach 511 of fitting module 51, the fit approach 531 of fitting module 53, can basis
represent the vector representation of Grey Point,
represent the vector representation of black color dots,
for the vector representation of merge coefficient.Fitting module 52 is identical with the structure of fitting module 51, but the different in kind of the matrix unit in the identical position of k-space matrix, and in fitting module 52, in the position of black color dots and Grey Point and fitting module 51, the position of black color dots and Grey Point is just in time exchanged, now,
represent the vector representation of black color dots,
represent the vector representation of Grey Point, because black color dots and Grey Point are in same k-space, the pass of coil ties up in this k-space constant, and therefore black color dots is equal to the position relationship of Grey Point, namely utilizes the vector that formula (1) and formula (2) are tried to achieve
theoretical value identical, therefore fitting module 52 in the present embodiment with fitting module 51, the status of 53 is equal to, the fit approach 521 of fitting module 52.Owing to also comprising the fitting module of similar setting in the present embodiment, by fitting module 51,52,53 and other similar settings fitting module merge generate total fitting module, calculate to obtain described merge coefficient by described total fitting module.In the present embodiment, because formula (1) and formula (2) are overdetermined equation, because formula (1) is equal with the status of formula (2), (1) is described for example namely with the formula, formula
for overdetermined equation, therefore exist
inverse, have
wherein H represents conjugate transpose, finally obtains vector by formula (1) '
Fig. 6 is the easy sampling schematic diagram of the magnetic resonance parallel image acquisition method that the embodiment of the present invention provides, simultaneously with reference to Fig. 5, by fitting module in Fig. 5 51,52,53 are placed on data Layer 61,62 respectively, 63, data Layer 61,62,63 respectively compared to fitting module 51,52,53, just respectively by fitting module 51,52, the secondary series of the k-space matrix comprised in 53 and the 6th column skip and not shown.Suspension points in Fig. 6 represents more data reconstruction layer.Adopt
in conjunction with fit approach 64 in the present embodiment, wherein
corresponding data structure 65,66, represents total fitting module respectively.Because formula (3) is overdetermined equation, have
i.e. basis
merge coefficient can be calculated to obtain by total fitting module.
Utilize the merge coefficient of above-mentioned acquisition to fill up lack sampling k-space, filled up by white point, form full sampled k-space.Finally utilize described full k-space data parallel acquisition magnetic resonance image (MRI).
The present invention also provides a kind of method for reconstructing of magnetic resonance image (MRI), utilizes above-mentioned magnetic resonance parallel image acquisition method to obtain full sampled k-space data, is converted into the view data of image area, merges the view data of described each passage thus realize image reconstruction.Fig. 3 c utilizes the method for the embodiment of the present invention when speedup factor is 2, the image rebuild after filling up disappearance k-space data.With reference to Fig. 3 c, with do not utilize the embodiment of the present invention method parallel acquisition figure compared with (Fig. 3 a, Fig. 3 b), convolution artifact can be removed clean according to the magnetic resonance image (MRI) of above-mentioned full k-space data parallel acquisition by the embodiment of the present invention.
Fig. 7 a to be speedup factor be 4 the method without traditional GRAPPA fill up, the image after directly utilizing the k-space of missing data to rebuild; Fig. 7 b to be speedup factor be 4 the method through traditional GRAPPA merge the image rebuild after disappearance k-space data filled up by coil; Fig. 7 c utilizes the method for the embodiment of the present invention when speedup factor is 4, the image rebuild after filling up disappearance k-space data.See Fig. 7 a, Fig. 7 b, Fig. 7 c, when speedup factor becomes large, 7a, 7b occurs that the situation of convolution artifact is more obvious, and these convolution artifacts can be removed by the magnetic resonance image (MRI) of the method parallel acquisition provided by the embodiment of the present invention, see 7c, the effect of removal is more obvious.
Obviously, those skilled in the art can carry out various change and modification to invention and not depart from the spirit and scope of the present invention.Like this, if these amendments of the present invention and modification belong within the scope of the claims in the present invention and equivalent technologies thereof, then the present invention is also intended to comprise these change and modification.